Please use this identifier to cite or link to this item:
Title: Diversidade, distribuição espacial, modelagem e manejo de um fragmento florestal em Moçambique com predominância de Androstachys johnsonii Prain.
Other Titles: Diversity, spatial distribution, modeling, and management of a forest fragment in Mozambique, with a predominance of Androstachys johnsonii Prain.
Authors: Calegario, Natalino
Acerbi Junior, Fausto Weimar
Barbosa, Bruno Henrique Groenner
Oliveira, Marcelo Silva de
Mendonça, Adriano Ribeiro de
Keywords: Fitossociologia
Redes neurais artificiais
Modelos não lineares
Manejo florestal
Artificial neural networks
Nonlinear models
Forest management
Issue Date: 1-Mar-2021
Publisher: Universidade Federal de Lavras
Citation: TUZINE, M. S. Diversidade, distribuição espacial, modelagem e manejo de um fragmento florestal em Moçambique com predominância de Androstachys johnsonii Prain. 2018. 105 p. Tese (Doutorado em Engenharia Florestal) – Universidade Federal de Lavras, Lavras, 2021.
Abstract: Inventory is an essential tool for forest management and operational management. However, this technique requires some additional tools to provide better estimates in data modeling, such as phytosociological analysis, geostatistics, hypsometric relationship, diametric distribution, and management prescription. Androstachys johnsonii Prain (Euphorbiaceae) is an endemic species in southern Africa widely exploited and threatened in southern Mozambique. This paper's objective was to assess the performance of spatial models, artificial neural networks, and nonlinear regression models to estimate biometric variables from forest inventory data. We installed 79 temporary Sample Units (SU) of 0.2 ha each and measured the trees with a diameter at breast height greater than 10 cm. We recorded 4978, representing 41 species and 17 families. Fabaceae was the dominant family, with 14 species, followed by Anacardiaceae, Combretaceae, Euphorbiaceae, and Loganiaceae, with three species each. The average basal area was relatively low (15.72 m²ha-1), accompanied by low levels of diversity of Shannon-Wiener species (H'=1.08) and uniformity index (J=0.61), indicating the occurrence of a monodominance. The geostatistical analysis revealed a gradient of biometric variables, increasing from the southern to the northern regions. The neural networks showed better performance for estimating total height, with an error of less than 2.4 m. However, the inclusion of covariants also improved the estimates, obtaining an error equal to 2.5 m. The biexponential model showed a better performance than the one generally used (Meyer–Liocourt), thus being recommended for unbalanced forests.
Appears in Collections:Engenharia Florestal - Doutorado (Teses)

Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.